Add setup_chat_format for adding new special tokens to model for training chat models
#1242
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
There was a problem hiding this comment.
Thanks a lot ! I left some early comments, in addition can you:
- document its usage in the SFTTrainer docs?
- and also add a slow test here: https://github.com/huggingface/trl/blob/main/tests/slow/test_sft_slow.py#L36 with an end-to-end example
Thanks!
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
lapp0
pushed a commit
to lapp0/trl
that referenced
this pull request
May 10, 2024
…aining chat models (huggingface#1242) * first draft * 64 * sourabs suggestion * wip tests * make style happy * add check * docstring * fix docstring * Update tests/test_model_utils.py Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com> * move tests * add todo for abstract class * make style happy * add slow tests and imports * add documentation * sft_trainer.mdx aktualisieren Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com> --------- Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
This PR adds a new
utilmethodsetup_chat_format, which automatically defines thechat_templatefor a tokenizer, adds special tokens, resizes the model embedding layer (optional to a multiple of 64)It also introduces the
ChatMlSpecialTokensdataclass, which is used in thesetup_chat_format. This will make it easy to extend to different formats in the future, e.g., llama, but for now, we only addchatml.Open Discussions
Should we add more dummy tokens to the tokenizer when the embedding layer is extended to a multiple of x? This can lead to downstream issues with llama.cpp